A method for predicting rolling mill hourly output based on nearest neighbor clustering

Abstract

The application discloses a kind of based on nearest neighbor clustering forecast rolling mill machine hour output method, comprising: obtaining the real-time rolling data of each material;Based on rolling time, the interval time, rhythm time and machine hour output of each material are calculated, and machine hour output information table is established;Based on material steel grade, material weight and product specification setting classification condition, using nearest neighbor clustering algorithm based on classification condition real-time to material is classified, and rolling mill machine hour output grading table is established;Real-time update each kind of product specification rolling mill interval time, rhythm time and machine hour output data and establish rolling mill machine hour output gear information log table;Obtain the information of material to be rolled, and based on each table, unknown rolling mill machine hour output is predicted.The application method is according to the characteristics of many rolling mill rolling product specifications, using the method of nearest neighbor clustering to integrate and cluster within the range of difference allowed specification, facilitate production forecast and production arrangement etc.Work is carried out, effectively improve production efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic control of rolling steel, and in particular to a method for predicting rolling mill hourly output based on nearest neighbor clustering. BACKGROUND

[0002] In steel production, the rolling mill hourly output refers to the weight of the material processed by the rolling mill in a unit of time. The purpose of calculating the hourly output is to serve as an index for comparing and analyzing the work efficiency of the steel production equipment and to master the operation of the equipment. The hourly output is analyzed to predict the time required for production planning orders and to arrange production and processes. However, the hourly output of different materials varies significantly, and factors such as adjustment of production processes and equipment wear cause the hourly output of the same product to change over time. In recent years, with the upgrading and transformation of steel production equipment processes and the diversification of demand for steel products in social production and life, the classification of steel products has increased, and the previous classification method no longer meets the current production situation. It is increasingly difficult for production personnel to control the time arrangement required to complete production orders, making it increasingly urgent to accurately classify and predict the hourly output.

[0003] The clustering method is to divide data samples according to the similarity of the data, so that samples with high similarity are classified into the same cluster, and samples with high difference are classified into different clusters. The prior art uses the nearest neighbor clustering technique to mine data information, which helps to improve the data processing capability and running efficiency of the algorithm program. If the clustering method can be used to study and predict the hourly output of steel production, it can meet the needs of actual steel production for production rhythm evaluation and efficient production arrangement, and has important theoretical research significance and engineering application value.

[0004] Therefore, there is a need in the prior art for a method for predicting the rolling mill hourly output based on nearest neighbor clustering. SUMMARY

[0005] Therefore, the present application provides a method for predicting the rolling mill hourly output based on nearest neighbor clustering, which uses the nearest neighbor clustering method to integrate and cluster the product specifications within the allowed range of differences according to the characteristics of the rolling mill rolling products, deeply mines the hidden information in the production data, facilitates production prediction and production arrangement, and effectively improves production efficiency.

[0006] To achieve the above purpose, the present application provides a method for predicting the rolling mill hourly output based on nearest neighbor clustering, which comprises the following steps:

[0007] S1, obtaining real-time rolling data of each material, the real-time rolling data including material steel grade, material weight, product specification, rolling mode and rolling time;

[0008] S2, calculate the rolling time, interval time, rhythm time and mill hourly output of each piece of material based on the rolling time, and establish a mill hourly output information table;

[0009] S3, set classification conditions based on the material steel grade, material weight and product specification, use the nearest neighbor clustering algorithm to classify the material in real time based on the classification conditions, and establish a mill hourly output classification table;

[0010] S4, real-time update of mill rolling time, interval time, rhythm time and mill hourly output data of each product specification and establish a mill hourly output gear information log table;

[0011] S5, obtain the information of the material to be rolled, and predict the unknown mill hourly output based on the mill hourly output information table, the mill hourly output classification table and the mill hourly output gear information log table.

[0012] In some embodiments, in S3, the specific steps of establishing the mill hourly output classification table include:

[0013] S3-1: Set the classification conditions as steel grade, material weight and product specification, set the vector x i based on the material weight and product specification, set the cluster center queue Q, set the maximum distance D max , read each piece of material data in time sequence, take the first slab data and add it to the queue Q as the first cluster center Q 11 of the steel grade gz1;

[0014] S3-2: Read the second slab data, if gz2≠gz1, take the second slab data and add it to the queue Q as the first cluster center Q 21 of the steel grade gz2;

[0015] If gz2=gz1, calculate the distance D 21 between the vector x2=(w2,kd2,hd2) and the cluster center vector x1=(w1,kd1,hd1), if D 21 <D max , slab P2 is classified into the cluster center Q 11 ; if x 21 >x max , slab P2 becomes the new cluster center Q 12 of the steel grade gz1;

[0016] S3-3: Read the third slab data, if gz3≠gz2, gz3≠gz1, take the third slab data and add it to the queue Q as the first cluster center Q 31 of the steel grade gz3;

[0017] If gz3 = gz2, gz3≠ gz1, calculate the distance D 21 of vector x3 from cluster center Q 31 , if D 31 > D max , slab P3 becomes new cluster center Q 22 ; otherwise, slab P3 is classified into cluster center D 21 of steel grade gz2.

[0018] If gz3 = gz2 = gz1, and there are cluster centers Q 11 , Q 12 , calculate the distance D 11 , D 12 of vector x3 from cluster centers Q 31 , Q 32 , if D 31 > D max , D 32 > D max , slab P3 becomes new cluster center Q 13 of steel grade gz1; otherwise, slab P3 is classified into the nearest cluster center.

[0019] Continue in this way until all samples are classified and the classification results are established into the mill hourly production grading table.

[0020] In some embodiments, the measuring units of classified condition material weight w i , target width kd i and target thickness hd i are different from the classification division range, the classification division of material weight is w i ± w a , the classification division of target width is kd i ± kd a , and the classification division of target thickness is hd i ± hd a , the values are pre-processed before classification:

[0021] w i = w i / w a

[0022] kd i = kd i / kd a

[0023] hd i = hd i / hd a .

[0024] In some embodiments, in S1, the product specification includes target width kdi , target thickness hd i , rolling time includes rolling start time t j and rolling end time t j of each rolling stage.

[0025] In some embodiments, the rolling stage includes a primary rolling stage, a warm-up stage, and a secondary rolling stage.

[0026] In some embodiments, in S2,

[0027] The mill net rolling time T i1 The calculation method is to calculate the difference between the rolling end time and the rolling start time of each stage and record it as the stage rolling time, and sum all the stage rolling times to obtain T i1 ;

[0028] The interval time T i2 The calculation method is to calculate the difference between the rolling start time of each stage and the rolling end time of the adjacent previous stage, and then sum the differences to obtain T i2 ;

[0029] The rhythm time T i The calculation method is:

[0030] T i = T i1 + T i2

[0031] The mill output Pro i The calculation method is:

[0032] Pro i = w i / T i .

[0033] In some embodiments, in S4, after the mill completes the processing of a slab, the slab's current mill net rolling time, interval time, rhythm time, and mill output data are calculated, and the historical sub-grade data are classified. In response to changes in the corresponding product specification mill output data, the mill output grade information log table records the update time and updates the product specification mill information.

[0034] In some embodiments, in S5, obtaining the material to be rolled information includes obtaining the material steel grade gz b , the material weight w b , the product target width kd b , the product target thickness hd b As screening information, the mill output sub-grade table is queried for the same steel grade and the most similar mill output sub-grade information, which is used as the prediction information of the unknown mill output.

[0035] In some embodiments, when the rolling mill machine time yield is queried, the same steel grade and the most similar specification are included.

[0036] In the rolling mill machine time yield grading table, the material steel grade gz is screened i =gz b The grading data is obtained by taking the material weight w of each rolling mill machine time yield grading data i , the target width kd i , and the target thickness hd i , the vector x i =(w i ,kd i ,hd i ) is calculated, and the distance D b between the vector x b =(w b ,kd b ,hd ib ) is calculated, and the rolling mill machine time yield grading corresponding to the minimum distance D ib is the predicted information.

[0037] In some embodiments, the rolling mill pure rolling time, interval time, rhythm time, and machine time yield data of each product specification rolling mill are updated in real time as product specification rolling mill samples, and the median of the samples is taken as the real-time statistical value of the product specification rolling mill pure rolling time, interval time, rhythm time, and machine time yield.

[0038] The present application has at least the following beneficial technical effects:

[0039] The present application provides a method for predicting rolling mill machine time yield based on nearest neighbor clustering, which calculates the rolling mill rhythm time and machine time yield of each material in the steel production, performs nearest neighbor clustering division on the slabs of the same steel grade and similar specification, and updates the predicted machine time yield information in real time, solving the defects of low calculation accuracy of rolling mill machine time yield, multiple product specifications, and difficulty in meeting the actual steel production requirements for production rhythm evaluation and efficient production arrangement. The rolling mill equipment rhythm time and machine time yield and other index information reflecting the production status of the equipment are displayed to the operator in real time, providing a basis for optimizing the production procedure and equipment state and timely judgment. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other embodiments according to these drawings without creative labor.

[0041] Figure 1 The schematic diagram of the method embodiment for predicting the rolling mill hourly output based on the nearest neighbor clustering provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0042] In order to make the objects, technical solutions and advantages of the present application clearer, the embodiments of the present application are further described in detail below with reference to the drawings.

[0043] In addition, the reference to "embodiments" in this document means that the specific features, structures or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily mean that it refers to the same embodiment, nor is it independent or alternative to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0044] The embodiment of the present application provides a method for predicting the rolling mill hourly output based on the nearest neighbor clustering, as shown in the figure. Figure 1 The flow chart of the method for predicting the rolling mill hourly output based on the nearest neighbor clustering includes the following steps:

[0045] S1: Obtain the real-time rolling data of each piece of material, and the real-time rolling data includes the material steel grade, the material weight, the product specification, the rolling mode and the rolling time;

[0046] Further, the real-time data of the material to be obtained includes: the material steel grade gz i , the material weight w i , the product specification (the target width kd i , the target thickness hd i ), the material rolling mode, the rolling start time t j of the material at a certain stage, the rolling end time t j of the material at a certain stage, the heating furnace number, the loading and discharging time, the controlled rolling code kz i , and the team information.

[0047] S2: Calculate the rolling mill pure rolling time, the interval time, the rhythm time and the hourly output of each piece of material based on the rolling time, and establish a rolling mill hourly output information table;

[0048] Further, the rolling mill pure rolling time T i1 , the interval time T i2 , the rhythm time T i and the hourly output Pro i of each piece of material are calculated, including:

[0049] The rolling mill pure rolling time T i1The calculation method is to calculate the difference between the rolling end time and the rolling start time of each stage and record it as the stage rolling time, and sum all the stage rolling times to obtain T i1 ;

[0050] Interval time T i2 The calculation method is to calculate the difference between the rolling start time of each stage and the rolling end time of the adjacent previous stage, and then sum the difference to obtain T i2 ;

[0051] The material rolling mill rhythm time is equal to the material rolling mill pure rolling time plus the interval time, and is calculated as follows:

[0052] T i = T i1 + T i2

[0053] The material machine hourly output is equal to the material weight divided by the material rhythm time, and is calculated as follows:

[0054] Pro i = w i / T i

[0055] In the embodiment of the application, the slab rolling mill hourly output board data information part is shown in Table 1.

[0056] Table 1 slab rolling mill hourly output data table

[0057]

[0058] S3: based on the material steel grade, material weight and product specification to set the classification condition, based on the classification condition to use the nearest neighbor clustering algorithm to classify the material in real time, and to establish a rolling mill hourly output classification table;

[0059] In some embodiments, the product specification numerical classification condition: material weight w i , target width kd i and target thickness hd i ;

[0060] In other embodiments, the product specification numerical classification condition further includes: material steel grade gz i , controlled rolling code kz i , heating furnace production mode jr i ;

[0061] In some embodiments, the material rolling mill hourly output data has great difference when the material is in single furnace operation production and double furnace operation production in the heating furnace, and the classification condition can add the material heating furnace production mode jr i , and is calculated as follows:

[0062] Obtaining the specified slab charging time T0, discharging time T0', intercepting the slab number and corresponding furnace number of the heating furnace within the range of T0-T0', if the furnace number is all single heating furnace, the specified slab heating furnace production mode is single furnace production, otherwise it is double furnace production;

[0063] In some embodiments, the classification condition material weight w i , target width kd i and target thickness hd i The units of measurement of the classification division are different, and the classification division of the material weight is w i ±w a , the classification division of the target width is kd i ±kd a , the classification division of the target thickness is hd i ±hd a , and the numerical value is processed before classification:

[0064] w i =w i / w a

[0065] kd i =kd i / kd a

[0066] hd i =hd i / hd a

[0067] In some embodiments, if the product specification numerical classification condition is set as material weight w i , target width kd i and target thickness hd i , the specific steps of real-time grading classification of materials of different products based on nearest neighbor clustering include:

[0068] S3-1: Set the classification condition as material steel grade gz i , material weight w i and product specification (target width kd i , target thickness hd i ), set the vector x i based on the material weight and the product specification, set the cluster center queue Q, set the maximum distance D max , read each piece of material data in time sequence, take the first piece of slab data and add it to the queue Q as the first cluster center Q 11 of steel grade gz1;

[0069] S3-2: Read the data of the second slab. If gz2≠gz1, add the data of the second slab to queue Q, and use it as the first cluster center Q for steel grade gz2. 21 ;

[0070] If gz2 = gz1, calculate the distance D between vector x2 = (w2, kd2, hd2) and cluster center vector x1 = (w1, kd1, hd1). 21 If D 21 = <D max Slab P2 is classified under cluster center Q. 11 If x 21 >x max Slab P2 becomes the new cluster center Q for steel grade gz1. 12 ;

[0071] S3-3: Read the data of the third slab. If gz3≠gz2 and gz3≠gz1, add the data of the third slab to queue Q as the first cluster center Q for steel grade gz3. 31 ;

[0072] If gz3 = gz2 and gz3 ≠ gz1, calculate the relationship between vector x3 and cluster center Q. 21 Distance D 31 If D 31 >D max Slab P3 becomes the new cluster center Q 22 Otherwise, classify slab P3 into cluster center D of steel grade gz2. 21 ;

[0073] If gz3 = gz2 = gz1, and there is a cluster center Q 11 Q 12 Calculate the relationship between vector x3 and cluster center Q. 11 Q 12 Distance D 31 D 32 If D 31 >D max D 32 >D max Slab P3 became the new cluster center for steel grade gz1. 13 Otherwise, classify slab P3 into the nearest cluster center;

[0074] This process continues until all samples are classified and a rolling mill hourly output classification table is established based on the classification results.

[0075] In other implementations, if the specification numerical classification condition is set to the material steel grade gz i Controlled rolling code kz i heating furnace production modei , material weight w i , target width kd i , and target thickness hd i , the specific steps of real-time grading classification of different specifications of materials based on nearest neighbor clustering include:

[0076] S3-1': set the classification condition as material steel grade gz i , control rolling code kz i , heating furnace production mode jr i , material weight w i , target width kd i , and target thickness hd i , set the cluster center queue Q, set the maximum distance D max , read each piece of material data in time sequence, take the first piece of slab data add to the queue Q as the first cluster center Q corresponding to (gz1, kz1, jr1) 1111 ;

[0077] S3-2': read the second piece of slab data, if gz2≠gz1 or kz2≠kz1 or jr2≠jr1, take the second piece of slab data and add it to the queue Q as the first cluster center Q of (gz2, kz2, jr2) 2221 ;

[0078] If gz2=gz1, kz2=kz1, jr2=jr1, calculate the distance D of vector x2=(w2, kd2, hd2) and cluster center vector x1=(w1, kd1, hd1) 21 , if D 21 <D max , slab P2 is classified into cluster center Q 1111 ; if x 21 >x max , slab P2 becomes the new cluster center Q corresponding to (gz1, kz1, jr1) 1112 ;

[0079] S3-3': read the third piece of slab data, if gz3=gz2=gz1, kz3=kz2=kz1, jr3=jr2=jr1, and there is cluster center Q 1111 , Q 1112 , calculate the distance D of vector x3 and cluster center Q 1111 , Q 1112 31 , D 32 , if D 31 >D max , D 32 >D max ​, slab P3 becomes the new cluster center Q corresponding to (gz1, kz1, jr1) 1113 ; otherwise, slab P3 is classified into the nearest cluster center;

[0080] If gz3 = gz2, gz3≠gz1, kz3=kz2=kz1, jr3=jr2=jr1, calculate the distance D of vector x3 and cluster center Q 2221 31 , if D 31 > D max , slab P3 becomes the new cluster center Q corresponding to (gz2, kz2, jr2) 2222 ; otherwise, slab P3 is classified into the cluster center Q 2221 ;

[0081] If gz3≠gz2, gz3≠gz1, kz3=kz2=kz1, jr3=jr2=jr1, take the third slab data added into the queue Q as the first cluster center Q corresponding to (gz3, kz3, jr3) 3331 ;

[0082] Continue in this way until all samples are classified.

[0083] Wherein, pg1 is the product code corresponding to the steel grade, product width, thickness, slab weight, the numbering rule is pg i = gz i *hd i *kd i *w i *kz i *jr i ; [T i ] is the rolling mill rhythm time instance sample queue, the rolling mill rhythm time of the material classified into the corresponding product is added into the queue every time a material is read, the maximum length of the queue is 600 records, and the record entered the sample queue the earliest is deleted when it exceeds; T i Take the median of the rolling mill rhythm time in [T i ] sample queue as the rolling mill rhythm time statistical value; [T i1 ], [T i2 ] are the rolling mill interval time and pure rolling time instance sample queues, T i1 , T i2 are the rolling mill interval time and pure rolling time statistical values, the rules are the same as the rolling mill rhythm time; [Pro i ] is the product rolling mill hourly output instance sample queue, the rolling mill hourly output of the material classified into the corresponding product is added into the queue every time a material is read, the maximum length of the queue is 600 records, and the record entered the queue the earliest is deleted when it exceeds; Pro i ​Take [Pro i ] The median of the rolling mill hourly output of the product is taken as the statistical value of the rolling mill hourly output.

[0084] In the embodiment of the present application, the partial results of the real-time grading classification of different products based on the nearest neighbor clustering are shown in Table 2.

[0085] Table 2 Rolling mill hourly output product grading table

[0086]

[0087] S4: Real-time update of rolling mill pure rolling time, interval time, rhythm time and rolling mill hourly output data of each product, and establishment of rolling mill hourly output grade information log table;

[0088] In some embodiments, after the rolling mill completes the processing of a slab, the slab's current rolling mill pure rolling time, interval time, rhythm time and rolling mill hourly output data are calculated, and the historical grading data are compared and classified. When the corresponding product rolling mill hourly output Pro i changes, the update time is recorded, and the product information is updated.

[0089] In the embodiment of the present application, the partial results of the rolling mill hourly output product grading table are shown in Table 3.

[0090] Table 3 Rolling mill hourly output product grading table

[0091]

[0092] S5: Obtain the rolling material information, and predict the unknown rolling mill hourly output based on the rolling mill hourly output information table, the rolling mill hourly output grading table and the rolling mill hourly output grade information log table;

[0093] In some embodiments, the rolling mill hourly output information table, the rolling mill hourly output grading table and the rolling mill hourly output grade information log table are obtained based on steps S1-S4. According to the screening conditions of the material steel grade gz b , the material weight w b , the product target width kd b , the product target thickness hd b , the rolling mill hourly output grading table is queried to obtain the rolling mill hourly output grade information of the same steel grade and the most similar specification as the prediction information of the unknown rolling mill hourly output.

[0094] In some embodiments, the query of the rolling mill hourly output grade information of the same steel grade and the most similar specification includes:

[0095] In the rolling mill hourly output grading table, the material steel grade gz i = gz bGear data, taking the material weight w of each mill hourly output gear data i Target width kd i Target thickness hd i Calculate vector x i =(w i ,kd i ,hd i ) and vector x b =(w b ,kd b ,hd b ) distance D ib , the mill hourly output gear corresponding to min{D ib} is the predicted information.

[0096] In the embodiment of the present application, by a method for predicting mill hourly output based on nearest neighbor clustering, the mill rhythm time and mill hourly output of each material in steel production are calculated, the slabs of similar specifications of the same steel grade are classified by nearest neighbor clustering, and the predicted mill hourly output information is updated in real time, so that the defects of low calculation accuracy of mill hourly output, many product specifications, and difficulty in meeting the needs of actual steel production for production rhythm evaluation and efficient production arrangement in the prior art are solved, the mill equipment rhythm time and mill hourly output and other index information reflecting the production status of the equipment are displayed to the operator in real time, and the basis for optimizing the production schedule and the equipment state is provided.

[0097] It should be particularly pointed out that each step in each embodiment of the above-mentioned method for predicting mill hourly output based on nearest neighbor clustering can be crossed, replaced, added, deleted, and therefore, these reasonable permutations and combinations of the method for predicting mill hourly output based on nearest neighbor clustering should also belong to the protection scope of the present application, and the protection scope of the present application should not be limited to the embodiments.

[0098] Finally, it should be pointed out that those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by a computer program to instruct related hardware, and the program of the method for predicting mill hourly output based on nearest neighbor clustering can be stored in a computer readable storage medium, and when the program is executed, the program can include the processes of the above-mentioned embodiments. The storage medium of the program can be a disc, an optical disc, a read-only memory (ROM), a random access memory (RAM), or the like. The above-mentioned computer program embodiments can achieve the same or similar effects as any of the above-mentioned method embodiments.

[0099] Moreover, the method disclosed by the embodiments of the present application can also be implemented as a computer program executed by a processor, which can be stored in a computer readable storage medium. When the computer program is executed by the processor, the above-mentioned functions defined in the method disclosed by the embodiments of the present application are implemented.

[0100] In addition, the above-mentioned method steps and system units can also be implemented by using a controller and a computer readable storage medium for storing a computer program enabling the controller to implement the above-mentioned steps or unit functions.

[0101] The above is the exemplary embodiments disclosed by the present application, but it should be noted that various changes and modifications can be made without departing from the scope of the embodiments disclosed by the present application defined by the claims. The functions, steps and / or actions of the method claims according to the disclosed embodiments described herein need not be performed in any particular order. Furthermore, although the elements of the disclosed embodiments can be described or claimed in individual forms, unless explicitly restricted, they can also be implemented in combination with each other.

[0102] It should be understood that, as used herein, the singular forms "a", "an" and "the" are intended to include plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising", when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0103] The above-mentioned embodiment number of the embodiments of the present application is only for description, not representing the advantages or disadvantages of the embodiments.

[0104] Those skilled in the art should understand that the above discussion of any embodiment is only exemplary, and is not intended to imply that the scope of the embodiments disclosed by the present application (including claims) is limited to these examples; under the idea of the embodiments of the present application, the above embodiments or technical features in different embodiments can also be combined, and there are many other changes of different aspects of the embodiments of the present application as above. In order to be brief, they are not provided in details. Therefore, any omission, modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A method for predicting rolling mill hourly output based on nearest neighbor clustering, characterized in that, include: S1. Obtain real-time rolling data for each piece of material, including the steel grade, weight, product specifications, rolling mode, and rolling time. S2. Calculate the pure rolling time, interval time, rhythm time and machine output of each piece of material based on the rolling time, and establish a rolling mill output information table. S3. Set classification conditions based on the steel grade, weight, and product specifications of the material. Use the nearest neighbor clustering algorithm to classify the material in real time based on the classification conditions and establish a rolling mill hourly output classification table. S4. Real-time update of pure rolling time, interval time, rhythm time and machine hour output data for each specification of rolling mill and establish a rolling mill hour output level information log table. S5. Obtain information on the material to be rolled, and predict the unknown mill hourly output based on the mill hourly output information table, mill hourly output grading table, and mill hourly output grading information log table. In S2, the pure rolling time T of the rolling mill i1 The calculation method is as follows: calculate the difference between the end time and the start time of rolling in each stage and record it as the stage rolling time. Summate the rolling times of all stages to obtain T. i1 ; The interval time T i2 The calculation method is as follows: calculate the difference between the start time of each rolling stage and the end time of the nearest previous rolling stage, and then sum the differences to obtain T. i2 ; The rhythm time T i The calculation method is as follows: T i =T i1 +T i2 The machine hour output Pro i The calculation method is as follows: Pro i =w i / T i ; In S3, the specific steps for establishing a rolling mill hourly output grading table include: S3-1: Set the classification condition to material steel type gz i Material weight w i and product specifications, the product specifications including target width kd i and target thickness hd i Vector x is set based on material weight and product specifications. i Set up a cluster center queue Q and a maximum distance D. max Read the data of each material in chronological order, and add the data of the first slab to queue Q, which will serve as the first cluster center Q for steel grade gz1. 11 ; S3-2: Read the data of the second slab. If gz2≠gz1, add the data of the second slab to queue Q, and use it as the first cluster center Q for steel grade gz2. 21 ; If gz2 = gz1, calculate the distance D between vector x2 = (w2, kd2, hd2) and cluster center vector x1 = (w1, kd1, hd1). 21 If D 21 = <D max Slab P2 is classified under cluster center Q. 11 If D 21 >D max Slab P2 becomes the new cluster center Q for steel grade gz1. 12 ; S3-3: Read the data of the third slab. If gz3≠gz2 and gz3≠gz1, add the data of the third slab to queue Q as the first cluster center Q for steel grade gz3. 31 ; If gz3 = gz2 and gz3 ≠ gz1, calculate the relationship between vector x3 and cluster center Q. 21 Distance D 31 If D 31 >D max Slab P3 becomes the new cluster center Q 22 Otherwise, classify slab P3 into cluster center D of steel grade gz2. 21 ; If gz3 = gz2 = gz1, and there is a cluster center Q 11 Q 12 Calculate the relationship between vector x3 and cluster center Q. 11 Q 12 Distance D 31 D 32 If D 31 >D max D 32 >D max Slab P3 became the new cluster center for steel grade gz1. 13 Otherwise, classify slab P3 into the nearest cluster center; This process continues until all samples have been classified and a rolling mill hourly output classification table has been established based on the classification results. In S4, after the rolling mill completes the processing of a slab, it calculates the pure rolling time, interval time, rhythm time and machine output data of the slab in this rolling mill, compares it with historical grade data for classification, and responds to changes in the machine output data of the corresponding specification rolling mill. The rolling mill output grade information log table records the update time and updates the specification rolling mill information.

2. The method for predicting rolling mill hourly output based on nearest neighbor clustering according to claim 1, characterized in that, Classification criteria: material weight w i Target width kd i and target thickness hd i The units of measurement and classification ranges differ; the classification of material weight is as follows: w i ±w a The target width is classified into kd i ±kd a The target thickness is classified into HD. i ±hd a Numerical preprocessing before classification: In i =in i / In a book i =kd i / kd a hd i =hd i / hd a 。 3. The method for predicting rolling mill hourly output based on nearest neighbor clustering according to claim 1, characterized in that, In S1, the product specifications include the target width kd. i Target thickness hd i The rolling time includes the rolling start time t of each rolling stage. j and the rolling end time t of each rolling stage j '.

4. The method for predicting rolling mill hourly output based on nearest neighbor clustering according to claim 3, characterized in that, The rolling process includes a primary rolling stage, a warming stage, and a secondary rolling stage.

5. The method for predicting rolling mill hourly output based on nearest neighbor clustering according to claim 1, characterized in that, In S5, obtaining information about the material to be rolled includes obtaining the steel grade (gz). b Material weight w b Product target width kd b Product target thickness hd b As filtering information, the hourly output level of the rolling mill with the same steel grade and the most similar specifications is retrieved from the rolling mill hourly output level table, and used as prediction information for the unknown rolling mill hourly output.

6. The method for predicting rolling mill hourly output based on nearest neighbor clustering according to claim 5, characterized in that, The information on the hourly output level of the rolling mill with the same steel grade and most similar specifications found includes: Material steel grade gz selected in the rolling mill hourly output classification table i =gz b Gear data, taking the material weight w from the hourly output gear data of each rolling mill. i Target width kd i Target thickness hd i Calculate vector x i =(w i ,kd i ,hd i ) and vector x b =(w b ,kd b ,hd b Distance D ib ,min{D ib The corresponding rolling mill hourly output level is the desired prediction information.

7. The method for predicting rolling mill hourly output based on nearest neighbor clustering according to claim 1, characterized in that, For each type of rolling mill, the pure rolling time, interval time, rhythm time, and machine output data are updated in real time as a sample of the rolling mill of the specified type. The median of the sample is taken as the real-time statistical value of the pure rolling time, interval time, rhythm time, and machine output of the rolling mill of the specified type.

Images